考虑数据不均衡的居民用户负荷曲线分类方法

由于用户用电行为的多样性和随机性,负荷数据存在分布不均衡的问题,传统负荷曲线分类方法在处理不均衡数据时性能较差,为此,提出一种改进K-means与LSTM（long short term memory network）-CNN（convolutional neural network）分类模型结合的负荷曲线分类方法。首先,为提升K-means算法对不均衡数据的聚类效果,基于密度峰值聚类算法（density peaks clustering,DPC）思想,提出一种相对k近邻密度峰值初始聚类中心选取方法（related k-nearest neighbor density peaks,RKDP）,将其作为K-means算法的初始中心进行聚类;其次,为提高RKDP_K-means处理高维负荷数据的性能,采用LSTM自编码器进行特征降维后再聚类（LSTM-auto-encoder RKDP_K-means,LARK）获得精准类别标签;最后,基于LSTM和CNN网络分别提取负荷特征构建负荷曲线分类模型,实现对大规模负荷曲线的分类。算例表明在大规模负荷曲线分类时,相比于LARK算法,本文所提方法轮廓系数指标提升29.7%,效率提升3.46倍,具有良好的负荷曲线分类效果。     

基于曲线拟合的毫米波雷达安装角度校准方法

毫米波雷达的安装角度校准是雷达正常使用并与摄像头进行数据融合的重要前提,雷达安装角度偏 转过大会导致雷达数据与摄像头数据融合失败,影响高级驾驶辅助系统(Advanced Driving Assistance System,ADAS) 的正常使用。文中提出一种基于曲线拟合的毫米波雷达安装角度校准方法,当车辆在道路上行驶时辅以车辆输入 的车速和偏航角信息,通过2000 个静止点进行曲线拟合得到雷达需要补偿的角度。相较于选择有护栏的道路进行 绕行和在标定场地部署角反射器进行安装角度校准的方法,这种方法适用的场景种类更多并且校准时间从15 分钟 以上缩短为5 分钟以下。经过实验验证,在花费更短时间完成校准后,校准精度与其它自校准方法相同为±5°。     

基于时频域深度网络的海面小目标特征检测

为了提升海杂波背景下小目标探测性能,本文提出一种基于时频域深度网络的特征检测方法。首先,将观测向量转换为归一化时频图（Normalized Time-Frequency Graph, NTFG）,实现海杂波抑制。在时频域,建立海杂波、含正多普勒偏移目标回波、含负多普勒偏移目标回波的三分类问题,精细化目标落在主杂波带内外的不同特性。其次,引入Inception-ResNet V2深度网络作为特征提取器,自主学习不同类别在NTFG上的深层差异性,并将差异性浓缩为一个2D特征向量。然后,在2D特征空间中,设计具有引导的三次样条曲线,获得虚警可控的判决区域,实现异常检测。最后,IPIX实测数据验证了所提算法的性能优势,能深入挖掘时频域的特性。     

Hierarchical and Hybrid Cell Load Balancing in 5G Heterogeneous Mobile Networks

In Fifth Generation (5G) Heterogeneous Mobile Networks (HetNets), deploying dense small cell networks makes user association more challenging. The process of collecting cell load information from the User Equipments (UEs) and broadcasting the feedback message involves significant overhead and time complexity. Moreover, the UEs may not know the optimum cell to reselect, satisfying its data rate requirements. In order to overcome these drawbacks, in this paper, we propose to design an Hierarchical and Hybrid Cell Load Balancing (HHCLB) technique using Selective Handoff. In this technique, the UEs of each cell are grouped into clusters depending on their proximity distance. Each cluster contains a cluster controller (CC) which is in charge of determining the intra-cell load and redirecting the cell-reselection request of a UE. If the data rate of any UE in a cluster becomes less than its required rate, then the cell reselection process is performed. By simulation results, it is shown that load balancing can be done proactively (implicitly) by the CCs when the load is unbalanced or can be done on demand (explicitly) when a UE send a request for cell reselection. In the case of Macro cells, HHCLB attains 71% higher throughput for low load scenario and 59% higher throughput for high load scenario. Similarly, in the case of Femto cells, HHCLB attains 19% higher throughput for low load scenario and 27% higher throughput for high load scenario.     

Secure and Light Weight Elliptic Curve Cipher Suites in SSL/TLS

In the current circumstance, e-commerce through an online banking system plays a significant role. Customers may either buy goods from E-Commerce websites or use online banking to move money to other accounts. When a user participates in these types of behaviors, their sensitive information is sent to an untrustworthy network. As a consequence, when transmitting data from an internal browser to an external E-commerce web server using the cryptographic protocol SSL/TLS, the E-commerce web server ensures the security of the user’s data. The user should be pleased with the confidentiality, authentication, and authenticity properties of the SSL/TLS on both the user’s web browser and the remote E-commerce web server. E-Commerce web servers should choose the best SSL/TLS cipher suites for negotiating the user in order to attain such optimistic scenarios, as the cipher suite used in SSL/TLS plays an important role in securing E-Commerce web servers. The paper primarily focuses on analyzing the SSL/TLS cipher and elliptic curves. The paper also recommends the best elliptic curve cipher suites for E-Commerce and online banking servers, based on their power consumption, handshake execution time, and key exchange and signature verification time.     

The Double Edge Sword Based Distributed Executor Service

Scalability is one of the most important quality attribute of software-intensive systems, because it maintains an effective performance parallel to the large fluctuating and sometimes unpredictable workload. In order to achieve scalability, thread pool system (TPS) (which is also known as executor service) has been used extensively as a middleware service in software-intensive systems. TPS optimization is a challenging problem that determines the optimal size of thread pool dynamically on runtime. In case of distributed-TPS (DTPS), another issue is the load balancing b/w available set of TPSs running at backend servers. Existing DTPSs are overloaded either due to an inappropriate TPS optimization strategy at backend servers or improper load balancing scheme that cannot quickly recover an overload. Consequently, the performance of software-intensive system is suffered. Thus, in this paper, we propose a new DTPS that follows the collaborative round robin load balancing that has the effect of a double-edge sword. On the one hand, it effectively performs the load balancing (in case of overload situation) among available TPSs by a fast overload recovery procedure that decelerates the load on the overloaded TPSs up to their capacities and shifts the remaining load towards other gracefully running TPSs. And on the other hand, its robust load deceleration technique which is applied to an overloaded TPS sets an appropriate upper bound of thread pool size, because the pool size in each TPS is kept equal to the request rate on it, hence dynamically optimizes TPS. We evaluated the results of the proposed system against state of the art DTPSs by a client-server based simulator and found that our system outperformed by sustaining smaller response times.     

Graphics Evolutionary Computations in Higher Order Parametric Bezier Curves

This work demonstrates in practical terms the evolutionary concepts and computational applications of Parametric Curves. Specific cases were drawn from higher order parametric Bezier curves of degrees 2 and above. Bezier curves find real life applications in diverse areas of Engineering and Computer Science, such as computer graphics, robotics, animations, virtual reality, among others. Some of the evolutionary issues explored in this work are in the areas of parametric equations derivations, proof of related theorems, first and second order calculus related computations, among others. A Practical case is demonstrated using a graphical design, physical hand sketching, and programmatic implementation of two opposite-faced handless cups, all evolved using quadratic Bezier curves. The actual drawing was realized using web graphics canvas programming based on HTML 5 and JavaScript. This work will no doubt find relevance in computational researches in the areas of graphics, web programming, automated theorem proofs, robotic motions, among others.     

Secured Data Storage Using Deduplication in Cloud Computing Based on Elliptic Curve Cryptography

The tremendous development of cloud computing with related technologies is an unexpected one. However, centralized cloud storage faces few challenges such as latency, storage, and packet drop in the network. Cloud storage gets more attention due to its huge data storage and ensures the security of secret information. Most of the developments in cloud storage have been positive except better cost model and effectiveness, but still data leakage in security are billion-dollar questions to consumers. Traditional data security techniques are usually based on cryptographic methods, but these approaches may not be able to withstand an attack from the cloud server's interior. So, we suggest a model called multi-layer storage (MLS) based on security using elliptical curve cryptography (ECC). The suggested model focuses on the significance of cloud storage along with data protection and removing duplicates at the initial level. Based on divide and combine methodologies, the data are divided into three parts. Here, the first two portions of data are stored in the local system and fog nodes to secure the data using the encoding and decoding technique. The other part of the encrypted data is saved in the cloud. The viability of our model has been tested by research in terms of safety measures and test evaluation, and it is truly a powerful complement to existing methods in cloud storage.     

基于双重注意力机制和GRU网络的短期负荷预测模型

电力负荷预测对电力系统的部署、规划和运行影响重大,但目前各输入特征对电网负荷情况影响的程度不稳定,且递归神经网络捕获负荷数据的长期记忆能力差,导致预测精度下降。提出一种基于双重注意力机制和GRU网络的预测新模型,利用特征注意力机制自主分析历史信息与输入特征间的关联关系,提取重要特征,并通过时序注意力机制自主选取GRU网络中关键时间点的历史信息,提升较长时间段预测效果的稳定性。在3个公开数据集上的实验结果表明,该模型在预测精度指标上表现良好,对比SVR、KPCA-ELM、DBN、GRU、Attention-GRU、CNN-LSTM、Attention-CNN-GRU模型预测精度分别提高了2.47、1.14、1.93、1.37、1.04、0.74、0.41个百分点。     

基于径向曲线与支持向量回归的颅骨修复方法

颅骨修复技术是对有缺损的颅骨补全对应的缺损部分,进而实现颅骨形状的完整性。针对高维颅骨数据,采用径向曲线来表示颅骨几何特征,结合最小二乘支持向量回归的方法构建颅骨修复模型。提取完整的三维颅骨模型的径向曲线,将其分为已有径向曲线和缺失径向曲线两部分作为训练样本,采用最小二乘支持向量回归统计模型复原出待修复颅骨的缺失径向曲线,进而合并生成待修复颅骨的完整径向曲线,通过迭代最近点算法将合并的颅骨径向曲线与颅骨统计模型进行匹配生成完整的三维颅骨模型。实验结果表明,该方法的平均误差达到6.834×10^-3,比主成分分析方法降低2.90倍,具有更好的修复效果。